Robust and Energy Efficient Service Provider Networks

This thesis focuses on the energy efficiency of service provider networks. It examines methodologies for reducing the energy consumption as well as greening the operation of the network through the integration of renewable energy systems. It studies the impact of these techniques on network performance and robustness. In order to reduce energy consumption we focus on the problem of energy-aware network planning and traffic engineering. Utilizing the network criticality metric, from network science, and an optimization formulation we quantify the energy-robustness trade-off of green traffic flow assignments. We also develop a robust, dynamic lightpath assignment algorithm, which provides energy savings at low loads and an order of magnitude better blocking performance at high loads. We then examine the challenge of integrating renewable energy in core networks to minimize cost and emissions. We show that renewable energy systems can compromise network robustness and increase disruptions in the network because usage must 'follow the sun'. We study the integration of energy storage systems to mitigate these concerns and quantify the potential savings in energy and cost. We develop a distributed, energy management algorithm to make energy storage charging/discharging decisions as well as electricity purchases for a service provider. Finally we examine the problem of content caching in service provider networks. We study the energy consumption of a caching scheme designed for performance and propose energy-aware modifications. We conclude by studying the impact of over-the-top video-on-demand services like Youtube and Netflix on service provider access networks and propose methods to reduce energy consumption and network congestion.Ph.D